GPT Image 2 prompts

Outfit: a feminine casual summer outfit featuring a {argument name="top" default="yellow-and-white gingham camisole top"} in lightweight cotton fabric with thin spaghetti straps, a softly gathered bust, a lightly structured empire-to-waist transition, and a flared peplum-like babydoll hem, with premium fabric texture, realistic garment construction, clean stitching, natural seam tension, believable fabric weight, subtle wrinkles, realistic drape, refined edge finishing, visible seams, real folds, well-made clothing details, high-quality material feel, and detailed garment realism; paired with {argument name="bottoms" default="light-blue washed denim short shorts"} with visible seams, natural fading, clean finishing, believable denim weight, subtle wrinkles, and realistic garment construction; finished with {argument name="shoes" default="ivory-to-light-beige chunky platform sport sandals"} with black trim, broad fabric or canvas-like straps, round metal hardware details, and a black rubber sole; add a random mix of delicate feminine minimal accessories in soft gold-tone or pearl-tone that naturally match the outfit; not costume-like, not cheap-looking.

上传一张人像作为人物参考，把原图中的人物替换成参考人像，保留原图的动作、表情、眼神、姿势、构图和鱼眼镜头效果。人物为黑色长卷发，五官自然，表情灵动，不要有生硬换脸感。

Outfit: a sporty feminine casual summer outfit featuring a {argument name="top" default="fitted white short-sleeve cotton jersey baby tee"} with a clean crew neckline, subtle contrast piping around the collar, a tiny generalized logo-like chest detail, premium fabric texture, realistic garment construction, clean stitching, natural seam tension, believable cotton jersey weight, subtle wrinkles, refined edge finishing, visible seams, and detailed garment realism; paired with a {argument name="skirt" default="high-waisted light periwinkle-blue flared mini skirt"} in lightweight cotton-blend twill with a softly gathered waistband, crisp pleat-like volume, realistic drape, real folds, clean hem finishing, believable fabric weight, high-quality material feel, and well-made clothing details; finished with white crew socks featuring black horizontal stripe details and a small generalized text-style print, plus {argument name="shoes" default="chunky white-and-gray athletic sneakers"} with mesh panels, leather-like overlays, detailed stitching, layered rubber sole construction, and realistic footwear materials; add a random mix of delicate clean casual accessories in white-tone, silver-tone, or minimal sporty feminine style that naturally match the outfit; not costume-like, not cheap-looking.

Goal: Create a highly detailed fantasy illustration of {argument name="character name" default="The Moon Rabbit's Apothecary"}, a cozy lunar medicine shop carved into a stone arch on the surface of the moon.

Canvas: Horizontal 4:3 cinematic composition, viewed from slightly outside the shop looking inward and outward at the same time. The left two-thirds are a warm apothecary interior; the right third opens onto a cold moon landscape under a starry black sky.

Main subject: A realistic white rabbit apothecary with soft fur, tall upright ears, black eyes, and delicate paws, standing behind a weathered wooden counter. The rabbit wears layered pale teal-green and cream robes with subtle floral embroidery, like a refined East Asian herbalist. It is grinding medicine in a carved green ceramic mortar with a pestle, calm and focused.

Setting and layout: The shop is built into a rough gray stone moon-arch, with cracked plaster, aged wood shelves, and an arched opening framing the lunar surface. Behind the rabbit, fill the wall with old shelves of glass bottles, corked jars, tiny vials, books, herbs, and softly glowing powders. On the right, show a cratered moon plain with a winding path of folded paper cranes leading into the distance. Above the path hangs a glowing full Earth like a blue-white moon ornament, suspended from a chain with beads and a tassel.

Visible repeated elements to include: Show exactly 1 white rabbit apothecary, 1 mortar and pestle, 1 hanging Earth ornament, 1 vertical shop sign with East Asian calligraphy, 1 luminous globe lamp on the counter, 1 green vase of small white flowers, 1 small glass terrarium-like potion dome, and 8 folded paper cranes on the moon path. Include exactly 19 prominent labeled jars or tags with readable hand-lettered English text: “A Quiet Heart,” “Courage to Begin,” “Dreams That Bloom,” “Patience of the Moon,” “Words That Heal,” “A Path Unfolding,” “Kindness Returned,” “Joy in Little Things,” “Clarity of Spirit,” “Love That Stays,” “Starry Inspiration,” “Peace in Small Moments,” “Wisdom in Stillness,” “Good Fortune,” “Loss of Stars,” “Gentle Strength,” “Memories that Warm,” “For Always,” and “Hope.”

Style: Magical realism, storybook fantasy, ultra-detailed painterly realism, soft moonlit atmosphere, warm candlelit interior contrasted with cool silver lunar light outside. Use aged teal, ivory, gray stone, soft gold, and moon-silver colors. Materials should feel tactile: dusty glass, cork stoppers, old paper labels, carved ceramic, embroidered fabric, rough moon dust, and weathered wood.

Lighting: Warm glows from potion bottles and the counter lamp illuminate the rabbit and shelves; cold blue-white rim light from the hanging Earth and moon surface lights the right side. Add tiny stars, gentle bloom, dust motes, and a quiet dreamy mood.

Constraints: No modern objects, no humans, no cartoon style, no harsh neon colors, no extra animals, no watermark. Keep the composition serene, intricate, and enchanting, with readable labels where specified.

Goal: Create a polished square AI innovator profile infographic about {argument name="character name" default="Thibault Sottiaux"}, styled like a modern tech magazine hero card.

Canvas: Square 1:1 layout, white background, deep cobalt blue and bright orange accent color palette, thin rounded blue border, subtle circuit traces along the bottom edge, dotted decoration and abstract network nodes in the corners. Use crisp vector layout combined with a soft watercolor portrait.

Layout: Two-column composition. The left side contains a large watercolor-style head-and-shoulders portrait at the top and a career timeline card below it. The right side contains the oversized name headline at the top, followed by a short affiliation line, then three rounded information cards stacked in a balanced grid. Add a footer label centered at the bottom.

Hero portrait: On the upper left, show a friendly smiling adult man with short brown side-parted hair, light skin, light stubble, and a black hoodie, rendered as a semi-realistic watercolor illustration with blue ink splashes behind him and thin orbital arc lines around the portrait.

Headline and top text: Large bold uppercase title in dark blue reading {argument name="headline text" default="THIBAULT SOTTIAUX"}. Under it, place a thin orange rule and the affiliation line: “OpenAI • Core Products Lead • Codex”, with orange dot separators.

Cards and content: Use exactly 5 rounded content cards, each with a blue outline, clear icon badge, and dark-blue section heading.

1. “Who He Is” card: icon of a user silhouette in a blue circle. Body text: “Thibault ‘Tibo’ Sottiaux is a Belgian software engineer and AI product leader at OpenAI. He leads Codex and helps drive core product work focused on AI systems that assist real software engineering. Before OpenAI, he worked at Google DeepMind, where he built AI/ML workflow infrastructure and later led human data work for Gemini.”

2. “Career Timeline” card: calendar icon. Include exactly 6 timeline entries with connected blue circles: “2009–2012: B.Sc., Computational & Applied Mathematics, UCLouvain”; “2012–2014: M.Sc., Computational & Applied Mathematics, UCLouvain”; “2015–2018: Google — software engineering roles”; “2018–2024: Google DeepMind — ML workflows and research infrastructure”; “2023–2024: Gemini Human Data Lead”; “2024–present: OpenAI — Codex leadership and core product work”.

3. “Key Contributions” card: orange lightbulb icon. Include exactly 5 contribution rows, each with a small square line icon on the left and an orange bullet before bold blue label text: “Codex Leadership: Leads OpenAI’s software engineering agent and its product direction.”; “Agentic Engineering: Helps build systems that understand codebases, plan tasks, write code, and review changes.”; “Gemini Experience: At DeepMind, contributed to the infrastructure and human-data systems behind Gemini.”; “Product + Platform Integration: Bridges research, product, and platform execution across AI tools.”; “Real-World Utility: Focuses on practical agentic workflows that help people do meaningful work.” The 5 left-side icons are, in order: code brackets, brain, sparkle, puzzle piece, and user-with-code.

4. “Why He Matters” card: star icon. Include exactly 4 checkmark bullets: “Connects frontier AI to everyday developer productivity.”; “Helps turn agentic AI into useful tools that do real work.”; “Combines engineering depth with product execution.”; “Influences how AI coding systems evolve into broader work agents.”

5. “Legacy” card: classical column icon. Include a short paragraph: “Thibault Sottiaux represents a new generation of AI leaders who unite research, engineering, and product thinking. Through Codex and broader core-product leadership at OpenAI, he is helping shape the future of agentic software development and practical AI systems.” Add a decorative blue 3D microchip illustration to the lower right of this card, with the OpenAI knot symbol on the chip, glowing circuit lines radiating outward, and a blue wireframe brain/cloud above it.

Footer: Centered letter-spaced footer text reading “AI INNOVATORS SERIES”, with two small orange dots on each side.

Visual style: Clean editorial infographic, premium SaaS/AI branding, strong hierarchy, rounded cards, thin blue dividers, dark navy typography, orange highlights, minimal shadows, high-resolution print-ready finish.

Constraints: Keep all text legible and spelled correctly. Use exactly 5 cards, exactly 6 timeline entries, exactly 5 key contribution rows, and exactly 4 why-it-matters bullets. Do not add extra people, logos besides the OpenAI knot on the chip, watermarks, or social media UI.

Create a square hip-hop album cover on a vivid mustard-yellow background with a worn vintage print texture, subtle grunge speckles, and distressed pale edges like an old record sleeve. Center the design around a motocross rider on a red, black, and white Yamaha dirt bike doing a hard turn in the lower half of the image, angled slightly toward the viewer, with dirt and mud spraying from the rear tire to the lower left. The rider wears all-black motocross gear, black gloves, black boots, and a surreal glossy blue mask with a long nose and round cartoonish eyes, giving the image a meme-like, uncanny streetwear energy. Typography: place exactly 2 main title text elements above the rider: small black gothic blackletter text reading {argument name="artist name" default="toka"} centered near the top, and huge bold black condensed block letters reading {argument name="album title" default="YAMAHA"} spanning nearly the full width behind the rider. Add exactly 1 explicit-content label in the bottom-right corner reading "PARENTAL ADVISORY EXPLICIT CONTENT" in a small black-and-yellow sticker style. Include visible motorcycle branding details: exactly 3 small Yamaha marks or labels on the bike, including one on the front number plate, one on the front fork/side area, and one near the rear swingarm. Composition should feel like a gritty rap mixtape cover combined with motocross action photography, high contrast, sharp subject cutout, minimal background, no extra characters, no extra vehicles, no additional large text.

Create a candid full-body lifestyle photograph inside a cozy, busy coffee shop. The main subject is a young {argument name="woman description" default="East Asian woman in her early 20s with soft natural makeup, fair skin, and dark shoulder-length hair in a loose low ponytail with airy bangs"} standing near the café counter, slightly turned in profile and gazing upward thoughtfully as if waiting patiently. She holds a clear plastic cup of iced latte with a straw in one hand, while her other hand is tucked casually into the pocket of her relaxed-fit high-waisted cream trousers. She wears an oversized {argument name="shirt color and fabric" default="sage green linen"} button-up shirt with sleeves loosely rolled to the elbows, wide-leg cream trousers, clean white sneakers, and a large minimalist black leather shoulder tote hanging from one shoulder. The café interior should feel warm and authentic: dark wooden counter and cabinetry on the left, glass pastry display, chalkboard menu boards with small unreadable writing above the counter, shelves with cups and coffee items, warm pendant lights, exposed dark ceiling beams, concrete floor, wooden chairs and small round tables on the right, and a bright window toward the back. Include exactly 7 visible background customers: one woman in a black shirt and blue jeans at the counter on the left, one partially hidden café worker or customer behind the subject, one person in a white shirt facing away near the counter, one woman in black near the middle, one woman in a white T-shirt standing by the service area, and two people near the back window. Use natural documentary photography, realistic proportions, soft ambient café lighting, subtle film-like colors, shallow-to-moderate depth of field, vertical 3:4 composition, eye-level camera, no posed fashion-shoot feeling, no readable branding, no watermark, and no extra text. Mood: {argument name="mood" default="quiet calm, comfort, and unplanned everyday beauty"}.

Goal: Create a high-end fantasy character design sheet showing exactly 8 full-body costume concepts for {argument name="fantasy faction" default="ornate nomadic mystics and swamp occultists"}, arranged as a clean 2-row by 4-column lineup.

Canvas: Wide landscape canvas, approximately 4:3 aspect ratio, divided into 8 equal vertical character panels with thin dark separators. Neutral gray-beige studio backdrop in every panel, soft floor shadows, no text, no logos, no UI.

Layout: Each panel contains one centered standing figure, full body visible, posed like a museum-quality concept art turnaround. Top row has 4 figures; bottom row has 4 figures. Keep the characters distinct but unified by extremely detailed materials, ancient ritual fashion, layered robes, weathered metalwork, and handcrafted ornamentation.

Subject details: Depict exactly 8 discrete characters:
1. Top-left: A masked desert pilgrim warrior wearing curved ram horns, an orange-brown face mask, layered ivory wrappings, white ribbons blowing outward, blue-and-gold striped lower robes, sandals, leather straps, small armor pieces, and a large brass vessel or urn strapped on the back.
2. Top-second: A heavier robed ram-horned figure seen in a partial back/side pose, with a massive decorated cylindrical gold vessel on the back, carved horned headpiece, layered cream and blue robes, embroidered panels, dangling cloth strips, belts, and tarnished metal ornaments.
3. Top-third: A dark swamp shaman in a ragged black-green cloak made of leaves, moss, roots, and torn fabric; wide flat brim hat covered in vegetation, antler-like twigs rising from it, trailing black ribbons, hidden face, ornate organic chest details, mysterious silhouette.
4. Top-right: A more elaborate black-robed occult priest with a very wide spiked hat, hanging moss and cords, horn-like protrusions, dense rope necklaces, layered tattered foliage cloak, black and olive palette, holding a gnarled staff with dangling strips and charms.
5. Bottom-left: A blue celestial noble or astrologer wearing a deep cobalt velvet quilted robe with gold embroidery, fur cuffs and boots, pale face, hood, and a huge golden astrolabe ring structure behind the body like a halo, with circular mechanical arcs.
6. Bottom-second: A richly dressed astronomer-mage in dark navy robes with gold trim, bulky boots, fur edging, coin-like medallions, ornate chest plates, and an enormous armillary sphere headdress with multiple metal orbital rings; holding a large crescent-shaped astronomical instrument at the side.
7. Bottom-third: An insectoid or birdlike mechanical pilgrim with thin legs, beaked helmet, rounded blue-black shell helmet, layered cream-and-gold apron cloths, small backpack, curved arms, and one raised hand holding a small ornate lantern or censer.
8. Bottom-right: A taller insectoid reliquary carrier with a long narrow body, thin jointed legs, domed ribbed helmet, vertical hanging red-and-cream cloth banners, miniature architectural shrine structures strapped to the back, and a tall staff topped with stacked lantern-like relics and dangling chains.

Visual style: Ultra-detailed dark fantasy concept art, couture costume design, cinematic realism, painterly digital rendering, muted earthy palette with accents of tarnished gold, deep blue, ivory, black, olive green, and rust. Intricate embroidery, beadwork, filigree, rope, metal vessels, astrolabe machinery, ceremonial objects, torn fabric, and aged patina. Lighting is soft studio illumination from upper front, realistic shadows, crisp silhouettes, high texture detail.

Constraints: Use exactly 8 characters and exactly 8 panels. No captions or labels. No modern clothing, weapons, buildings, background scenery, or extra characters. The sheet should look like professional production concept art for {argument name="media type" default="a dark fantasy film"}.

Goal: Create a vertical 2-column by 5-row collage containing exactly 10 high-angle outdoor selfie photos of the same young woman, as if shot in rapid succession from an extended arm above her face.

Canvas: Tall vertical image, approximately 9:16 aspect ratio, filled edge-to-edge with a seamless photo-grid collage. No borders, captions, logos, or extra text.

Layout: Exactly 10 rectangular panels arranged in 5 rows and 2 columns. Each panel shows the same subject from a slightly different pose and facial expression while keeping the same camera angle, outfit, lighting, and location. The camera is held above and slightly in front of her, looking down; her extended forearm is visible in most panels.

Subject details: A young woman with {argument name="hair color" default="long dark brown hair"}, pale skin, slim build, glossy makeup, soft red lips, and delicate wire-frame glasses. She wears a vivid {argument name="shirt color" default="royal blue"} oversized button-up shirt worn open at the chest over a dark low-cut top, beige shorts, a small necklace, and bracelets. Her hair is wind-tousled with strands crossing her face. She is crouching or kneeling on a gravel path.

Panel count and expressions: Show exactly 10 variations: 1. serious parted-lips gaze upward, 2. head tilted with soft gaze, 3. downward shy gaze through glasses, 4. slightly surprised open-mouth look, 5. dreamy half-lidded expression, 6. neutral direct gaze, 7. focused intense look, 8. bright smiling expression, 9. sultry side-tilted gaze, 10. gentle closed-mouth smile.

Environment and lighting: Outdoor gravel ground with scattered leaves and mottled tree shadows. Warm natural sunlight creates dappled highlights across the subject, clothing, hair, and ground. Background remains mostly textured gravel and foliage shadow, with no buildings or other people.

Visual style: Hyper-realistic smartphone photography, candid social-media selfie aesthetic, shallow natural depth, slightly wide-angle perspective, high detail skin and fabric texture, realistic shadows, subtle lens softness, natural color grading with strong blue clothing contrast.

Constraints: Keep the same woman and outfit in all 10 panels, maintain a cohesive selfie sequence, do not add text, avoid watermarking, avoid extra panels, and keep the collage grid clearly visible.

Goal: Create a polished Japanese corporate strategy slide about the purpose and future vision of a project using accumulated data from an insurance-safety network and BI tools.

Canvas: 16:9 horizontal presentation slide, white background, navy and light-blue corporate palette, thin blue outlines, clean grid alignment, dense but readable Japanese business-slide typography.

Title: Large navy heading at top left: 「1.1. 本事業の目的と将来像」.

Top overview box: Directly under the title, add a white rectangular box with a thin navy border. On the left, include a dark navy pill label 「概要」. To the right, write: 「本事業では、保安ネットに蓄積されたデータをBIツール等で統計分析し、産業保安・製品安全行政における許認可・規制の確実な履行および蓄積データ毀損予防を実現するための調査を行う。」

Overall message band: Below the overview, add a pale blue horizontal band. On the left, a dark-blue arrow label 「全体メッセージ」. Main message text in bold navy: 「保安ネット×BIツール活用により、現状4課題は予兆検知と2大自動化の3将来像へ集約される。」

Main layout: Three columns with headings from left to right: 「目的」, 「現状課題」, 「将来像」. Use two horizontal rows separated by a dotted divider. Use arrows from current issues toward future vision cards.

Left column, exactly 2 dark navy purpose cards:
1. Card label 「目的1」 in a small white rounded pill, main text: 「産業保安・製品安全行政における許認可・規制の確実な履行」.
2. Card label 「目的2」 in a small white rounded pill, main text: 「蓄積データの毀損の恐れに係る予防」.

Center column, exactly 4 current issue items grouped into two pale-gray rounded containers, each item with a navy circular number badge:
1. 「画一的な規制の実施」 with smaller explanation 「認定事業者制度を除き、事業者の保安レベルに関わらず、一律の個別規制・許可届出・検査義務を課している。」
2. 「外部機関による統計分析業務の実施」 with smaller explanation 「関係課室の多くで委託事業者が事故情報の統計分析を実施しており、報告頻度は年1回に限定されている。」
3. 「データ集計に誤りが発生」 with smaller explanation 「第208回通常国会の高圧ガス保安法改正法案審議資料に、集計上の誤りが発生した。」
4. 「情報連携漏れのリスクがある」 with smaller explanation 「組織間連携をメール等で運用しているため、連携漏れや転記ミスによってDB不備が発生する恐れがある。」

Right column, exactly 3 future vision cards with blue outline and simple line icons:
1. Top large card connected from the first issue group by a thick navy arrow, icon of a magnifying glass with rising bar chart and warning triangle, text: 「事故・法令違反の予兆検知の実現」.
2. Middle card connected from the second issue group by a branching arrow, icon of a computer monitor/document with upload or transfer arrows, text: 「データ入力の自動化」.
3. Bottom card connected by the same branching arrow, icon of a database cylinder with gear, text: 「データ集計の自動化」.

Bottom message strip: Full-width white strip with thin blue border. On the left, a circular target icon. Text in bold navy: 「保安ネットの蓄積データをBIツールで可視化・分析することで、規制運用の高度化とデータ品質向上を同時に実現」.

Visual style: Japanese consulting deck aesthetic, crisp vector icons, consistent navy #003b8f / deep blue #001f5b accents, pale blue highlight bands, rounded cards, subtle gray dividers, no photos, no shadows beyond very light card depth. Keep all text sharp and legible, with the exact count of 2 purpose cards, 4 issue items, and 3 future vision cards. Use {argument name="project data source" default="保安ネット"}, {argument name="analysis tool" default="BIツール"}, {argument name="slide title" default="1.1. 本事業の目的と将来像"}, {argument name="overall message" default="保安ネット×BIツール活用により、現状4課題は予兆検知と2大自動化の3将来像へ集約される。"}, and {argument name="bottom message" default="保安ネットの蓄積データをBIツールで可視化・分析することで、規制運用の高度化とデータ品質向上を同時に実現"}.

Goal: Create a clean Japanese business infographic slide explaining the shift to assigning roles to AI, centered on the concept of an “AI employee.” Use {argument name="main headline" default="AIに役割を持たせる時代へ。"} as the large top headline.

Canvas: Vertical portrait infographic, approximately 3:4 aspect ratio, white background, thin light-gray border, navy blue accent color, black body text, rounded rectangular content blocks, minimalist line icons. The design should look like a polished LinkedIn/X business slide or proposal handout.

Layout: Use 6 stacked sections from top to bottom.

Section 1 — Header and cycle diagram: At the top, place the large bold headline. On the left, include three short Japanese explanatory lines about spending one week developing a personal AI editor-in-chief, repeatedly writing roles, assigning work, checking output, and rewriting roles. On the right, show a circular 4-step workflow with navy arrows and simple icons. Count exactly 4 workflow steps: 1) pencil icon labeled 「役割を書き直す」, 2) document icon labeled 「役割を書く」, 3) person-at-desk icon labeled 「仕事を任せる」, 4) magnifying glass icon labeled 「アウトプットを確認」.

Section 2 — Three company announcement cards: A rounded box with the heading {argument name="announcement heading" default="そうしたら今週、3社が「仕事をするAI」を発表しました。"}. Inside it, create exactly 3 horizontal company cards. Card 1: OpenAI logo-like knot icon, title 「OpenAI」, product 「ChatGPT Work」, short Japanese description about announcing a new AI work structure that autonomously handles complex tasks. Card 2: Anthropic stylized AI logo, title 「Anthropic」, product 「Claude Cowork」, short Japanese description about expanding Claude Cowork, where AI works as a member of the team. Card 3: xAI logo, title 「xAI」, product {argument name="xAI product name" default="Grok 4.5"}, short Japanese description about launching a new flagship strong in coding and agent tasks. Below the cards, place a dark navy banner with white text: 「3社とも、「答えるAI」ではなく「仕事をするAI」を出してきました。」.

Section 3 — Key message block: Split into two columns. Left column has large bold text: 「AIを使う時代から、AIに役割を持たせる時代へ。」 Right column has a paragraph explaining that the whole industry is moving in this direction, and that ChatGPT Work, AI employees, and AI Company are no longer unusual concepts. End with a bold navy question: {argument name="closing question" default="皆さんはもう、AI社員導入していますか？"}.

Section 4 — Definition of AI employee: Start with a navy pill heading: 「AI社員って何？」. Explain in Japanese that, simply put, it is an AI with a role. Include exactly 3 small role cards under 「例えば自分なら、」: 1) person icon labeled 「編集長（SNS・ブランド）」, 2) laptop/code icon labeled 「開発（コード）」, 3) document/search icon labeled 「秘書（調査・整理）」. Add a note that each is given a separate role. On the right, write that by giving AI roles like people have job roles, output becomes more stable.

Section 5 — How to make it: Add a navy pill heading 「どうやって作るの？」 and explain that the method is surprisingly simple: create a Claude Code company prompt and operate it. Show exactly 4 numbered steps in a horizontal process. Step 1: 「Claude Codeで company/フォルダを作る」 with a dark terminal-style file tree showing exactly 7 files: company/, mission.md, vision.md, values.md, role_editor.md, role_developer.md, role_secretary.md, rules.md. Step 2: 「各役割のルールや基準を Markdownで書く」 with bullets: 「ミッション」, 「ビジョン」, 「判断基準」, 「話し方」, 「役割の詳細など」. Step 3: 「Claude Codeにプロンプトを渡す」 with a black speech bubble containing Japanese text asking it to read the company folder and act as the SNS editor-in-chief. Step 4: 「仕事を任せて、改善を繰り返す」 with a mini circular loop showing person-at-desk, magnifying glass, and pencil icons labeled work, check, fix.

Section 6 — Bottom takeaway: A rounded footer strip with a navy lightbulb icon on the left and bold Japanese takeaway text: {argument name="footer takeaway" default="大事なのはプロンプトの上手さではなく、「役割の設計」です。"}. On the right, add a small people/team icon and a sentence explaining that if you assume AI is something to train like a human, it becomes an employee you can rely on.

Visual style: Japanese corporate infographic, crisp typography, generous spacing, subtle gray dividers, navy highlights, flat monochrome line icons, no photos, no gradients, no clutter. Keep all Japanese text legible and aligned. Do not add extra companies, extra steps, extra cards, or decorative elements beyond the specified counts.

Goal: Create a refined horizontal city good-morning poster for {argument name="city name" default="Shanghai"}, combining monumental macro flower photography with delicate editorial information layers and Chinese typography.

Canvas: Wide 16:9 poster, clean white and misty gray palette, high-key lighting, airy negative space, soft morning haze, premium archive-poster aesthetic.

Main visual: Use exactly 2 oversized white magnolia blossoms with real macro physical texture. Blossom 1 is cropped heavily from the upper-right edge, showing translucent white petals, a green stem base, and warm beige stamens. Blossom 2 is cropped from the lower-left edge, showing large white petals and part of the flower center. A single long green magnolia stem runs diagonally from the upper-right flower down toward the lower-left flower, creating a suspended connection and gentle visual tension. Keep the flowers enormous, almost monumental, with visible petal veins and soft backlit translucency.

Background: A faint Shanghai skyline sits low across the bottom third in pale fog, with the Oriental Pearl Tower near the center-right and other skyscrapers barely visible. Add a calm water reflection beneath the skyline. The background must remain subtle and desaturated so the flowers and typography dominate.

Layout and text elements: Include exactly 7 distinct information elements: 1) a small rounded capsule at the top-left reading “城市 / 上海” with a tiny purple dot on the right; 2) a thin horizontal keyword line on the left reading “关键词 / 海风 外滩 露光 玉兰” ending with a small circular node; 3) a large central vertical black Chinese headline reading “上海” above “早安”; 4) a purple handwritten script word “Shanghai” crossing diagonally through the central headline; 5) a white outlined quote card on the right with a folded-corner outline, purple quotation mark, the quote “早安，让风把灵感吹亮。”, a short purple divider line, and the date “2026-07-11”; 6) a slim vertical coordinate tag at far right reading “坐标 / SH-0711” with a vertical divider line and a small globe icon near the bottom; 7) a bottom-center copyright line reading “© 版权 / CITY POSTER ARCHIVE 2026”.

Typography and style: Use bold high-contrast black Chinese title characters, elegant Song/Ming-style Chinese text for small labels, and expressive purple cursive handwriting for the English city name. Accent color is vivid violet-purple, used only for the dot, script, quote mark, divider, and small right-side dot. Keep all spacing precise, minimal, and editorial.

Constraints: Do not add extra flowers, extra cards, people, vehicles, birds, logos, or watermarks. Maintain the exact two-blossom composition with top and bottom edge cropping, the diagonal stem connection, soft Shanghai skyline, and clean morning-poster atmosphere.

Create an ultra-realistic aerial cityscape photograph of {argument name="city" default="Guangzhou, China"} at golden-hour sunset, viewed from a high drone perspective over the Pearl River. The composition shows exactly 6 major focal elements: 1) the Canton Tower standing near the center in the hazy distance, 2) a tall dark glass office tower on the left with the low sun flaring through its upper floors, 3) a white curved-roof stadium or convention center in the lower middle, 4) the Pearl River running horizontally through the scene with bright gold reflections, 5) a large arched suspension bridge on the right crossing the river, and 6) sweeping elevated highways and ramps curving through the lower-right foreground with small cars and light trails. Surround these with dense modern skyscrapers, riverside greenery, multilane roads, and a warm atmospheric skyline fading into haze. Use {argument name="lighting" default="dramatic warm sunset backlight"}, orange and gold clouds, realistic reflections on glass buildings and water, cinematic HDR detail, sharp architectural realism, wide-angle 16:9 framing, high resolution, natural color grading, and a small Chinese watermark reading “豆包AI生成” in the bottom-right corner.

Goal: Create a clean Japanese product announcement presentation slide for {argument name="product name" default="GPT-5.6"}, with a futuristic AI/model-family visual system.

Canvas: 16:9 widescreen slide, white background, ample negative space, minimalist corporate tech style in navy blue and light blue. Use crisp vector graphics, subtle dotted orbital lines, and no photo texture.

Layout: Left side contains the typography block; right side contains a connected orbital diagram of exactly 3 model circles. Keep the composition airy, with most content in the upper and middle portions and a small page number at bottom right.

Text content: Main headline in very large bold navy Japanese text: “{argument name="headline text" default="GPT-5.6、ついに登場"}”. Under it, smaller bold black Japanese subtitle: “{argument name="subtitle text" default="答えるAIから、仕事を完成させるAIへ"}”. Below the subtitle on the left, add one small rounded rectangle date badge with thin blue outline and blue text: “{argument name="date text" default="2026.07"}”. Bottom-right page number: “01”.

Diagram details: Show exactly 3 connected circular model nodes, arranged horizontally from largest to smallest: 1) a large dark navy “Sol” circle near center-right with a white radial dot/starburst icon above the label “Sol”, surrounded by multiple thin concentric orbit rings, dotted arcs, and small blue/teal nodes; 2) a medium blue “Terra” circle to its right, connected by a dotted line with small solid blue connector dots, containing a white globe icon above the label “Terra”, also surrounded by a few thin orbital rings; 3) a small pale blue “Luna” circle further right, connected the same way, containing a white crescent moon and small star icon above the label “Luna”, with thin rings around it.

Visual style: Use deep navy for the headline and Sol node, medium cobalt for Terra, pale sky blue for Luna, and tiny teal accent dots. Typography should look like a modern Japanese sans-serif presentation deck, bold for the headline, clean and geometric for labels. Keep the orbital diagram precise, balanced, and technical, with thin strokes and small circular nodes.

Constraints: Use exactly the three visible model labels Sol, Terra, and Luna; include exactly one date badge and one page number; do not add extra paragraphs, logos, watermarks, icons outside the described diagram, or decorative background patterns.

Create a photorealistic vertical beachside portrait of {argument name="character name" default="a young East Asian woman"} kneeling on a soft bed beside an open ocean-view window on a bright tropical morning. She wears a wide frayed straw sunhat with long white lace ties, has {argument name="hair color" default="dark brown"} hair in a loose side braid with wispy bangs, and gazes calmly off to the right with a natural, sunlit expression. Her outfit is a delicate {argument name="dress style" default="pale pink floral sundress with thin straps and a deep gathered neckline"}, rendered as tasteful summer fashion. Add dappled shadows from the straw hat across her face, shoulders, arms, and dress, with glossy warm skin highlights. The bed has white sheets and a rumpled {argument name="blanket color" default="pink polka-dot blanket"}. Include exactly 2 cats resting on the bed in the lower left: one black-and-white cat partly behind her and one orange tabby sleeping in front. On the left edge, include a sheer white lace curtain with floral embroidery, softly backlit. Through the large window show turquoise ocean water, white surf, pale sand, a blue sky with puffy clouds, and a small palm-covered headland on the right. Use realistic DSLR-style photography, shallow-to-medium depth of field, crisp natural details, high dynamic range sunlight, relaxed tropical vacation mood, vertical 3:4 composition, no text, no watermark.

Create a high-key studio beauty portrait of a young East Asian woman, framed from upper torso to head, looking directly into the camera with a gentle closed-mouth smile. She has {argument name="hair color" default="dark brown"} long softly wavy hair styled half-up with airy curtain bangs and a few loose strands around the face, clear fair skin with natural rosy blush, glossy peach-pink lips, subtle eyeliner, long lashes, and warm brown eyes. Her left hand is raised gracefully to her cheek with relaxed fingers, creating a soft beauty-ad pose. She wears a simple {argument name="top color" default="white"} sleeveless sheer chiffon blouse over a camisole, minimal jewelry, no visible logos. Use a clean warm off-white studio background with abundant negative space on the right, soft diffused frontal lighting, very gentle shadows, smooth retouched skin, bright commercial Japanese cosmetics photography style, realistic proportions, shallow depth of field, crisp eyes, natural texture, polished post-processed look. Vertical square crop, face positioned slightly left of center, shoulders angled subtly, calm elegant atmosphere.

Create an ultra-realistic fisheye peephole view from inside a dark door, looking out at a cute young Japanese woman with {argument name="hair color" default="dark brown"} shoulder-length hair and soft bangs, leaning very close to the lens in a narrow apartment hallway. She has large glossy brown eyes, natural makeup, rosy cheeks, a slight gentle smile, and her face is enlarged by the wide-angle distortion. She wears a {argument name="hoodie color" default="mustard yellow"} hoodie with cream drawstrings, a simple dark bear face graphic on the chest, small dark paw-print details on one sleeve, and a plush bear-ear-like accessory visible behind her shoulder. Her left hand rests against the doorframe as she peers in curiously. The composition is circular, framed by a thick black peephole vignette and metallic lens rim, with strong barrel distortion and the subject centered but slightly tilted. Background: a plain, dim, narrow indoor corridor with off-white walls, gray floor, doors along the right side, fluorescent ceiling lights, and subtle grime and wear for realism. Lighting is low and cinematic, with soft highlights in her eyes, realistic skin texture, detailed hair strands, shallow depth of field, high-resolution photorealistic detail, no text, no watermark, no extra people.

Create a vertical cinematic fine-art portrait of a young East Asian woman in an elegant dark floral gown, set against a deep matte black background. The woman has {argument name="hair style" default="short wavy black bob with soft bangs"}, porcelain skin, delicate makeup, and a calm melancholic expression, looking down and to the right rather than at the viewer. Her shoulders are bare under sheer black tulle, with a lace-like black choker and embroidered blossoms across the bodice. Around and behind her, compose an ornate Japanese-inspired abstract collage: exactly 5 prominent white satin ribbon streams spiraling vertically through the image, 7 large visible cream or white flowers, 4 red or rust circular paper fan/disc motifs, 3 dark folding fan shapes, many fine gold botanical line drawings, thin looping gold arcs, scattered bronze ink splatters, antique beige paper textures, and layered dried-flower silhouettes. The main decorative column rises from the lower center to the upper left, while the portrait occupies the lower-right third. Use a luxurious dark palette of black, ivory, antique gold, sepia, muted red, and warm bronze. Lighting should be dramatic and soft, with the face and shoulder gently illuminated from the front-left, glossy highlights on skin, and surrounding collage elements fading into shadow. Style: ultra-detailed photorealistic fashion editorial mixed with Japanese paper collage, lacquerware elegance, wabi-sabi textures, high contrast, shallow depth of field, no text, no watermark, no border.

Goal: Create a clean Japanese economic infographic slide summarizing the corporate goods price index速報 for {argument name="report month" default="2026年6月"}.

Canvas: 16:9 horizontal presentation slide, white background with very pale blue accents, ample margins, polished corporate report design. Use navy as the main text color and bright medium blue for numerical percentages. Add a small gray page number “01” at the bottom right.

Layout: Large headline at the top left reading “企業物価指数、{argument name="report month" default="2026年6月"}速報”. Directly below it, a blue subtitle reading “国内物価は前年比+7.1%、輸入物価は円ベースで+29.7%”. Center the main content as exactly 3 rounded rectangular statistic cards in one horizontal row, each with a light blue border, subtle shadow, faint blue gradient fill, a circular icon area at top left, a bold Japanese category title, a thin divider line, and two metric rows. At the bottom, place a small friendly white-and-blue robot mascot on the lower left, pointing upward, and a centered conclusion sentence.

Statistic cards: Use exactly 3 cards. Card 1: factory icon, title “国内企業物価”, metric rows “前月比 +0.4%” and “前年比 +7.1%”. Card 2: cargo ship icon, title “輸出物価（円ベース）”, metric rows “前月比 +0.4%” and “前年比 +20.7%”. Card 3: globe and container/warehouse icon, title “輸入物価（円ベース）”, metric rows “前月比 +1.3%” and “前年比 +29.7%”. Make the percentages large and bright blue, aligned to the right of each row label.

Bottom text: Centered near the bottom in bold navy Japanese: “6月の企業物価は、国内・輸出・輸入で上昇が続いた”.

Visual style: Modern Japanese business presentation, crisp vector icons, rounded corners, soft shadows, balanced whitespace, consistent alignment. No charts or extra decorations beyond the 3 cards, the robot mascot, and page number. Keep all Japanese text legible and exactly as specified.

Goal: Create a vertical hand-drawn Japanese infographic on dotted notebook paper explaining that {argument name="AI model name" default="Muse"} changes illustrated infographic creation, positioning it as a third option for Japanese image generation.

Canvas: Portrait 4:5 layout, warm off-white cream paper background with a subtle dotted grid. Use a cute stationery scrapbook style with sketchy navy ink outlines, pink masking tape, beige tape, tiny sparkles, leaves, paperclip doodles, arrows, and hand-drawn border lines. Color palette: navy blue, dusty pink, cream, soft gray, and small red/blue chart accents.

Top header: A wide pale pink ribbon banner across the top with bold dark navy Japanese headline text: 「{argument name="headline text" default="Museで図解制作が変わる？"}」. Under it, centered red subtitle: 「{argument name="subtitle text" default="日本語の画像生成、第3の選択肢"}」.

Main layout: Three white sketch-outlined content cards on the left and center, plus an illustrated presenter character on the right. Use dashed pink arrows to connect the cards and suggest a workflow.

Card count: Exactly 3 cards.
1. Upper-left card: Pink tape label with bold navy text 「従来AI」. Below, two lines: 「文字が弱い」 and a speech bubble 「日本語も崩れる」. Illustration: a messy pile of crumpled papers with chaotic scribbles, exclamation marks, and a small chart sheet with an alert icon.
2. Center card: Pink tape label 「正確さ」. Below, two lines: 「図表を作る」 and 「コードと検索」. Illustration: a QR-code-like square, a magnifying glass, a bar chart with red and blue bars, a wrench icon, and a gear icon.
3. Lower-left card: Pink tape label 「自己修正」. Below, two lines: 「描き直せる」 and 「手間が軽くなる」. Illustration: circular refresh arrows, a small eraser, and a mini chart sheet with check marks.

Right side details: Add a checklist with exactly 3 checked items, written in English: “Nano Banana Pro”, “GPT Image 2”, and “Muse Image”. Add a vertical side note near the lower right reading “Nano Banana Pro” with two small empty checkboxes beside the margin.

Presenter character: On the right, draw a friendly Japanese businesswoman in simple anime/chibi infographic style, about one-third of the image height. She has {argument name="character hair" default="shoulder-length dark brown bob hair"}, black round glasses, a white button-up blouse with sleeves rolled up, a navy pencil skirt, and navy flats. She smiles and points left toward the central card.

Decorative element count: Include exactly 2 pieces of tape at the top corners, exactly 1 paperclip near the upper right, exactly 1 blue tape strip near the lower left, several small sparkles and doodle curls around the edges, and thin hand-drawn border strokes at the top and bottom.

Visual style: Clean but handmade, Japanese social-media explainer graphic, legible hand-lettered text, thick marker-like navy outlines, slight imperfections, soft pastel fills, no photorealism.

Constraints: Keep all Japanese text accurate and readable, avoid extra headings or extra cards, maintain the exact three-card structure and the exact three-item checklist, no watermark, no realistic photography.

Goal: Create a warm, photorealistic overhead tabletop scene of an open summer picture diary drawn in cute colored-pencil style, as if made by a child, featuring the same small anime girl character in two diary entries.

Canvas: Landscape 4:3 image, top-down view with slight natural perspective, cozy daylight on a wooden desk, high detail, realistic objects around the notebook, no digital UI or watermark.

Main subject: An open grid-paper notebook centered on the desk, showing exactly 2 illustrated diary pages. Each page has a large crayon-and-colored-pencil drawing at the top and neat handwritten Japanese diary text underneath. The character is a chibi anime girl named {argument name="character name" default="a cheerful blonde chibi girl"}, with short twin-tail hair, pale skin, blue-green eyes, a flower hair accessory, a patchwork colorful summer top, plaid skirt, patterned socks, and cute sneakers. Keep her consistent across both drawings.

Left diary page: Header date reads {argument name="left page date" default="7月20日（土）"}. The drawing shows the girl sitting on a wooden porch in daylight, happily holding a triangular slice of watermelon. Include a bright sunflower on the left, blue sky with fluffy clouds, a hanging wind chime near a wooden post, and a plate with exactly 2 watermelon wedges on the porch. Add a small red sun doodle near the date and a small smiling sun doodle near the lower right of the page. Under the drawing, write the diary text as {argument name="left diary text" default="おそとで　すいかを　たべたよ。\nあまくて　つめたくて　おいしかった！\nひまわりも　げんきだったよ。"}.

Right diary page: Header date reads {argument name="right page date" default="7月25日（木）"}. The drawing shows the same girl standing by a river at night watching a fireworks festival. Include a dark blue starry sky, exactly 5 large colorful fireworks bursts, reflections on the water, a distant town skyline with glowing windows, grass along the riverbank, and a small vertical festival sign reading 花火大会. Add exactly 2 small yellow star doodles near the top. Under the drawing, write the diary text as {argument name="right diary text" default="はなびたいかいに　いったよ。\nキラキラして　とても　きれいだった！\nドーンって　おとが　びっくりしたよ。"}.

Surrounding tabletop objects: Include exactly 4 loose crayons above the notebook, labeled by color in Japanese: peach, yellow, pink, and blue. Add a glass of iced tea on a woven coaster at upper left, exactly 1 cicada insect near it, exactly 2 seashells on the left side, a colored-pencil tin at lower left with visible pencils, a small blue pencil sharpener, pencil shavings, exactly 2 loose colored pencils near the bottom, a colorful fireworks crayon box near the bottom center, exactly 2 small toy figures near the bottom right, a folded towel at upper right with cute sea-creature embroidery and Japanese name text, a round paper fan at right with morning glory flowers, and a bowl at lower right containing watermelon slices.

Visual style: The diary drawings must look handmade with visible waxy crayon texture, uneven outlines, childlike proportions, soft pastel colors, and grid-paper texture. The surrounding desk and objects should be realistic, warm, tactile, and nostalgic, with shallow natural shadows and summer vacation atmosphere.

Constraints: Countable elements must match exactly: 2 diary pages, 2 appearances of the girl, 2 watermelon wedges on the left-page plate, 5 fireworks bursts, 2 yellow star doodles, 4 crayons above the notebook, 2 seashells, 2 loose colored pencils, and 2 small toy figures. Do not add extra diary pages, extra characters, or printed watermarks.

Create a wide 16:9 anime-style gag comic scene on a bright modern Japanese city sidewalk. The composition is split by a building corner: on the left, two chibi anime girls hide in the shadow of a gray concrete wall, and on the right, a third chibi anime girl has just slammed into a metal utility pole at high speed. Use clean cel-shaded illustration, crisp line art, vivid colors, exaggerated manga motion effects, and a playful slapstick tone.

Scene details: The background is an urban shopping street with glass storefronts, office buildings, vertical shop signs, a blue sky, and a long sidewalk receding in perspective. Place a tall cylindrical utility pole near the right foreground. The impact victim should be a petite chibi military girl with long wavy {argument name="hair color" default="pink-blonde"} hair, a black-and-red military cap, black tactical outfit with red accents, spiral dizzy eyes, open shocked mouth, and arms stretched forward as her face/body hits the pole. Add two large orange-and-black manga impact sound effects reading 「ドン!」, one near the upper impact and one near the lower right, plus a pink sound effect 「キュン」 beside the impact burst. Include orange starburst impact shapes behind her and strong horizontal speed lines trailing from left to right.

Left-side prank setup: Show exactly 2 hidden chibi girls. Girl 1, standing far left, is a short lavender-haired girl with glasses, a white naval-style cap, white-and-blue uniform, white boots, and a calm deadpan expression with crossed arms. Girl 2, standing center-left, is a cheerful short blonde girl in a black tactical outfit with gold trim and a small black cap, holding a fishing rod. The fishing line extends diagonally upward and then down toward the street, suspending exactly 1 small rectangular photo/card bait showing a dark portrait image. Give the blonde girl a mischievous smile as if luring the victim.

Speech bubbles: Add exactly 2 vertical Japanese speech bubbles on the left. The lavender-haired girl says 「スペックオーバーですね」. The blonde fishing girl says 「オーバースペックじゃね？」. Keep the text readable, black on white bubbles with thin black outlines.

Constraints: Keep the image comedic and dynamic, no gore or injuries beyond cartoon dizziness, no extra characters in the foreground. Maintain the 3-character count, the 2 speech bubbles, the 1 dangling photo bait, and the 3 visible sound-effect texts.

Create a photorealistic cinematic vertical selfie of a {argument name="character type" default="Japanese yokai fox-girl gyaru"} wandering through modern Tokyo at night in the rain. The subject is a young woman with long black hair, large realistic fox ears with tan inner fur, subtle fang-like charm, glossy makeup, small star stickers or glitter under one eye, and a playful wink. She holds up a V-sign near her cheek while extending her other arm toward the camera as if taking a close wide-angle smartphone selfie. She wears a Y2K gyaru-inspired outfit: cropped pale pink fitted top, dark cropped jacket with wet reflective fabric, and a dark pleated mini skirt. She stands beside a brightly lit convenience-store or arcade frontage with red vending machines visible through glass on the left, neon green and warm orange signage reflections, wet pavement, raindrops, and colorful Tokyo street bokeh in the background. Include one transparent umbrella above and behind her, speckled with raindrops and catching rainbow lens flares. Mood: cute, mischievous, urban supernatural nightlife. Style: ultra-detailed realistic anime-cosplay photography, 35mm wide-angle selfie perspective, shallow depth of field, wet reflections, cinematic teal-orange lighting, high contrast, soft skin detail, visible rain droplets, lens flare and light bloom. Composition: portrait 3:4, subject centered slightly right, face sharp, outstretched forearm large in the foreground, background softly blurred but recognizable as rainy Tokyo. Avoid extra characters, text, logos, distorted hands, duplicate ears, or overly cartoonish rendering.

Create a high-energy vertical anime concert illustration of an all-female gothic rock band performing live on a dark indoor stage. The image should feel like a dramatic festival moment, with intense purple, magenta, blue, and white spotlights, haze, lens glow, metal truss ceiling, confetti-like sparkles, and a packed audience in the foreground raising silhouetted hands in rock-sign gestures. Show exactly 4 visible band members: 1) the central lead singer and guitarist, a young woman with long flowing dark brown hair with pink highlights, amber eyes, black choker, layered necklaces, studded black leather jacket, black graphic top, plaid mini skirt, fishnet stockings, thigh garter, bracelets, chains, and lace-up boots, singing into a black microphone while playing a dark red electric guitar; 2) the left bassist, a cool-looking young woman with short lavender bob hair, black choker, black gothic outfit, plaid skirt, fishnets, holding a purple bass guitar near a microphone stand; 3) the rear drummer, a smaller young woman with dark hair behind a drum kit, smiling while playing, with a stylized white band logo on the bass drum; 4) the right keyboardist and backing vocalist, a young woman with dark twin-tail hair with pink accents, black gothic outfit and plaid skirt, singing into a microphone while standing behind a keyboard labeled Roland. The central performer should dominate the composition from head to thigh, placed slightly left of center, with the guitar neck rising diagonally upward. Use a polished Japanese anime style with detailed linework, glossy highlights, expressive faces, flowing hair, dramatic stage lighting, realistic instrument details, and a vivid live-performance atmosphere. Add a large abstract white band-logo graphic on the back wall behind the performers. Customizable details: band name/logo text {argument name="band name" default="stylized unreadable gothic logo"}, lead singer hair color {argument name="lead singer hair color" default="dark brown with pink highlights"}, lighting palette {argument name="lighting palette" default="purple, magenta, blue, and white"}, stage mood {argument name="stage mood" default="explosive live festival energy"}, keyboard brand label {argument name="keyboard brand label" default="Roland"}. Avoid photorealism, avoid extra band members, avoid readable unrelated text, and keep the image vertical with a close crowd perspective.